The purpose of this research is to evaluate the predictive accuracy of the STAR (Strategic Tool for Analysis of Risks) framework by comparing its health hazard risk classifications with actual emergency occurrences across selected countries in West and Central Africa over a three-year period.

The STAR tool does not generate numerical predictions of case counts. Its output is a composite risk score derived from four dimensions: Likelihood, Severity, Vulnerability, and Coping Capacity, all summed to produce a total score ranging from 4 to 25. This score is then classified into five categories: Very Low, Low, Moderate, High, and Very High. For visualization purposes, these categorical risk levels were converted into numeric codes (e.g., Risk 1 = Very Low, Risk 2 = Moderate, Risk 3 = High). These numeric codes were plotted alongside actual case counts in figures and tables to illustrate correlation trends. They do not represent predicted case numbers but rather the relative risk classification assigned by STAR.

As a retrospective observational study, it analyzes existing data from past events to identify patterns, correlations, or outcomes without manipulating variables or intervening in real time. In this context, the study examines historical health risk predictions made using the STAR Tool. It compares them with actual emergency occurrences across selected countries in West and Central Africa from 2022 to 2024. A mixed-methods approach combines quantitative analysis of outbreak data and a review of qualitative preparedness measures.

The study is conducted on health risk profiling carried out from January 2022 to December 2024 using STAR.

The study focuses on nine countries in West and Central Africa to evaluate the effectiveness of the STAR (Strategic Tool for Assessing Risks) in various contexts, specifically in West Africa (including Guinea, Senegal, Mali, Burkina Faso, and Nigeria) and Central Africa (comprising Gabon, Cameroon, the Republic of Congo, and the Central African Republic).

The selection ensures broad geographical and contextual representativeness across the two sub-regions: by including countries from both West Africa and Central Africa, the study considers the distinct epidemiological settings and regional differences in health systems and WHO Regional Office structures (AFRO). Taking into account the diverse environmental and security contexts, the sample encompasses a wide range of geographical and political environments, including coastal states (e.g., Gabon, Senegal, Nigeria), landlocked countries (e.g., Mali, Burkina Faso, Central African Republic), Sahelian zones (Mali, Burkina Faso), which face high levels of insecurity and humanitarian crises that directly affect health vulnerability, and forest and equatorial zones (e.g., Cameroon, Congo, Gabon), which have different disease profiles and environmental risks.

Regarding varying disease profiles, this diverse geographical selection ensures that the STAR tool’s predictive capabilities are tested across a full range of regionally common hazards, such as meningitis (seasonal in the Sahelian belt), cholera (linked to environmental and sanitation factors, often near coastlines or major rivers), and measles (related to immunization coverage in all settings). Additionally, including countries from both Francophone and Anglophone regions, such as Nigeria, helps evaluate the tool’s performance within different health administration models and levels of preparedness.

The outbreak data used in the study have been verified by combining standardized and independent official sources. The methodology explicitly addressed verification by implementing safeguards to ensure objectivity and scientific integrity of the findings, considering potential conflicts of interest. The outbreak data was sourced and verified using official and standardized documents and systems such as National Surveillance Data (obtained directly from the Ministries of Health), International Reports (outbreak reports and situation updates from WHO, Ministries of Health, and humanitarian platforms), Independent Data sources (national surveillance systems and WHO situation reports to confirm outbreak occurrences as a way to reduce bias), and official documentation from institutional repositories, government surveillance systems, and international health agencies with appropriate permissions. This triangulation process, which merges quantitative outbreak data from surveillance systems with qualitative assessments from official reports, served as a key measure to validate the actual occurrence of health emergencies.

The study uses a retrospective observational design. The analysis compares two main variables namely the predicted Risk Level and the actual occurrence. The comparison relies on interpreting whether a high-risk prediction visually corresponds to a high number of raw cases, rather than adjusting the case data for population size, reporting capacity, or other factors that normalization would account for. Triangulation (merging quantitative outbreak data from surveillance systems with qualitative assessments from official reports) was a key measure to validate the actual occurrence, which is a measure of verification rather than statistical normalization.

The study used a retrospective, observational, mixed-methods design to assess the correlation between STAR-predicted risk levels and actual outbreak occurrences across nine countries in West and Central Africa (2022–2024).

Identification of strengths and limitations in STAR’s predictive capacity, recommendations for improving risk assessment and preparedness planning, Evidence to support strategic resource allocation and early warning systems are the expected products of the research at the attention of WHO Regional Office for Africa, Ministries of Health, Humanitarian and development partners, and Emergency preparedness and response teams.

Each hazard is assessed using a composite risk score, which is the sum of four key dimensions:

Component	Description
Likelihood	Probability that the hazard will occur in each month.
Severity	Expected impact if the hazard occurs (e.g., mortality, morbidity).
Vulnerability	Susceptibility of the population (e.g., age, immunization status, poverty).
Coping capacity	Ability of the health system to respond (e.g., infrastructure, workforce).

The Strategic Tool for Assessing Risks (STAR) primarily categorizes the risk rather than giving actual numbers of predicted cases.

The Strategic Tool for Assessing Risks (STAR) provides categorical risk assessments rather than numerical case predictions. Its primary output is a composite risk score derived from four dimensions Likelihood, Severity, Vulnerability, and Coping Capacity—summed to produce a total score ranging from 4 to 25. This score is classified into five categories:

For ease of visualization in tables and figures, STAR’s descriptive risk categories were converted into numeric codes such as Risk 1 for Very Low; Risk 2 for Moderate and Risk 3 for High. This coding was applied only for graphical representation and does not alter STAR’s official classification system (very low, low, moderate, high, very high).

For meningitis, Predictions demonstrated the highest consistency with actual cases. Countries classified as Moderate to High risk generally experienced seasonal outbreaks, confirming STAR’s reliability for hazards with predictable drivers. As concerns measles: Correlation was variable. In some countries (e.g., Cameroon and Côte d’Ivoire), high-risk classifications aligned with major outbreaks. However, in others (e.g., Burkina Faso), STAR underestimated risk due to dynamic factors such as declining immunization coverage and service disruptions. Regarding cholera: Predictions were least reliable. Several countries classified as High risk reported zero cases, likely due to effective interventions or environmental conditions not captured by STAR. Conversely, Cameroon experienced a significant outbreak despite similar risk classification, indicating limitations in accounting for local variability.

STAR was highly effective for Meningitis because the disease’s risk factors are well-defined, highly seasonal, and predictable: Strong Seasonality: Meningitis outbreaks in West and Central Africa are strongly tied to the dry season (the “Meningitis Belt”) and predictable climate factors (low humidity, dust, wind). These drivers are relatively stable and can be captured by static models. The disease follows a highly consistent and understood epidemiological pattern across the region, making the Likelihood component of the STAR score inherently more reliable. Moreover, while interventions exist, the seasonal climatic drivers remain dominant and less susceptible to rapid, localized changes, allowing the tool’s assessment to hold true over the monitoring period.

Cholera predictions were the least reliable because the disease dynamics are highly sensitive to rapidly changing factors, particularly human interventions: There is a high sensitivity to preparedness because cholera risk is quickly and drastically mitigated by effective Water, Sanitation, and Hygiene (WASH) interventions, which are difficult for a static risk assessment tool like STAR to quantify accurately. For instance, the high-risk prediction for Côte d’Ivoire resulted in zero cases, interpreted as a success of proactive, effective preparedness measures that effectively negated the predicted threat. The tool accurately signaled the potential for risk, but the actual outcome was determined by response capacity. Additionally, it is observed that environmental variability with localized factors like rainfall, flooding, population displacement, and rapid urbanization affects contamination and transmission, introducing high variability that the tool struggles to capture in its composite score.

Measles predictions showed moderate but variable accuracy because the key driver, immunization coverage, is dynamic and subject to external shocks such as reliance on Dynamic Human Factors: Measles risk is fundamentally driven by immunity gaps in the population. These gaps are caused by constantly changing factors such as routine immunization rates, campaign effectiveness, and service disruption; factors like the COVID-19 pandemic or civil insecurity severely disrupt vaccination campaigns, leading to sudden, unpredictable drops in coverage. The STAR tool’s relatively static inputs often underestimated the risk because it did not fully account for these rapid, catastrophic service delivery collapses that fueled major outbreaks (e.g., the surge in cases in Burkina Faso).

In summary, the high predictability of Meningitis’s environmental drivers made it well-suited for STAR, while the dynamic and intervention-sensitive nature of Cholera and Measles introduced volatility that limited the tool’s raw predictive power. The study demonstrated the highest consistency between STAR’s predicted risk levels (typically Moderate to High in the Meningitis Belt countries) and actual emergency occurrences for meningitis (Figure 4) in Central African Republic. This strong correlation is largely attributed to the disease’s established, non-linear seasonality (peaks in the dry season) and established epidemiological patterns in the African meningitis belt. The results from countries like Burkina Faso (Table 2) affirm that STAR is highly reliable for hazards whose drivers (such as dry season climate) are relatively constant and measurable. This finding aligns with established public health literature, which consistently links meningitis outbreaks in Africa to specific climate and geographical factors (e.g., wind, dust, humidity) that make the hazard’s timing highly predictable for risk modeling (4). Cholera forecasts showed the least reliability, characterized by frequent overprediction and occasional underestimation of outbreak severity. The most significant discrepancy was the overprediction observed in countries like Côte d’Ivoire and Burkina Faso (Table 1), where a high predicted risk resulted in zero or near-zero actual cases. Conversely, the high predicted risk in Cameroon still underestimated the peak severity of the 2022 outbreak, which saw 1992 cases (Table 1) This weak correlation suggests that while STAR accurately captures vulnerability factors (e.g., poor sanitation, population density), it struggles to account for the dynamic impact of interventions and real-time environmental changes (5). Cholera dynamics are highly sensitive to Coping Capacity factors like prompt vaccination campaigns and effective Water, Sanitation, and Hygiene (WASH) improvements, which can rapidly decouple the predicted risk from the actual case counts, a challenge confirmed by recent mathematical modeling approaches (6). Measles predictions showed a moderate correlation but high variability, influenced heavily by changing immunization coverage and socio-political factors. Where predictions were accurate (e.g., Cameroon and Côte d’Ivoire), the high-risk classification was validated by major outbreaks with 10,404 and 2,154 cases, respectively (see Figure 2 and Tables 4, 5). Discrepancies arose due to underestimation of outbreaks stemming from low immunization rates and COVID-19-related service disruptions, as seen by the surge in cases in Burkina Faso in 2024. This finding directly reflects the long-term impact of the COVID-19 pandemic, which caused widespread disruption of routine immunization services and contributed to a substantial accumulation of susceptible populations (7). This drop in immunization coverage—a critical component of STAR’s Vulnerability score—was likely not fully captured by the static input data used in the STAR assessments, leading to a consistent underestimation of outbreak size and severity.

Regarding the role of the STAR Tool in Preparedness, it is observed that the cases of overprediction (e.g., zero cholera cases in Burkina Faso despite high risk, Table 3) serve as strong evidence that the Coping Capacity dimension of STAR can effectively mitigate the consequences of high intrinsic risk. The study suggests that STAR is most valuable as a strategic planning tool for hazard prioritization, rather than a precision forecasting instrument. Where STAR identified a high risk (Risk 3), it accurately signaled the need for increased preparedness, which in several cases (cholera control) appears to have successfully prevented the predicted emergency. This demonstrates a successful function of the STAR output: driving preventative action.

To improve accuracy, particularly for dynamic diseases like cholera and measles, the STAR framework requires enhanced integration with real-time, granular data sources related to climate and Environmental Data (to better model evolving conditions), vaccination Coverage Data (to track drops in herd immunity in real time) and cross-border coordination effectiveness (a key factor in the spread of measles and other pathogens).

Between 2022 and 2024, predictive models were assessed for their ability to forecast Cholera, Meningitis, and Measles outbreaks across African countries using the STAR tool, which assigns monthly risk scores based on likelihood, severity, vulnerability, and health system capacity (8, 9).

In Burkina Faso, cholera was predicted to be a moderate risk, but no cases occurred, suggesting either an overestimation or successful prevention. Meningitis predictions aligned with seasonal outbreaks, validating the model, while measles showed a mismatch moderate predicted risk versus a surge in early 2024 indicating the need for recalibration (10).

Regional cholera trends revealed seasonal peaks and disparities, with Guinea and Niger most affected. West Africa recorded over 399,000 cases and 7,000 deaths between 2022 and mid-2024, driven by rains and poor sanitation (11). Ebenezer et al. (6) employed machine learning to categorize countries based on outbreak patterns and estimate indicators such as R₀, thereby supporting the development of tailored interventions. Tang et al. (4) highlighted climate-sensitive factors air pollution, temperature, dust impacting meningitis in the Sahel, reinforcing climate-based predictive models for early warning. Charkhakan and Heravi (12) proposed an integrated risk framework combining probability, impact, and manageability to strengthen preparedness. In Burundi, predicted risks did not match actual cases (zero reported), possibly due to underreporting or surveillance gaps. Nipa and Allen (13) emphasized patch-specific modeling and seasonal calibration to improve prediction accuracy. In Cameroon, cholera predictions underestimated the 2022 outbreak (more than 1,700 cases), followed by a sharp decline. Meningitis remained low but persistent, while measles predictions were accurate, especially during the 2023 peak (14). Ebenezer et al. (6) applied a Bayesian framework and unsupervised learning to classify outbreak dynamics, thereby reinforcing the need for localized risk assessment in STAR. Bhavithra and Devi (7) developed a time-varying SEIR model for measles, supporting STAR’s objectives of risk identification, contextual analysis, and response planning. In CAR, cholera was predicted as a moderate risk but absent, likely due to WASH interventions. Meningitis aligned with seasonal dry periods, and measles predictions were validated by a 2023 peak (15). In Côte d’Ivoire, measles surged in 2024, confirming high-risk predictions, while cholera (high-risk forecast) did not occur. Meningitis remained moderate with seasonal peaks, aligning with STAR’s classifications. In the Republic of the Congo, WHO reported a suspected triple outbreak of typhoid, shigellosis, and cholera in 2023. Despite cholera being predicted as a very high risk, only 21 confirmed cases occurred, indicating a localized outbreak and gaps in surveillance. The response included multisectoral coordination and community engagement, reflecting STAR principles (16, 17). Overall, STAR proved useful for hazard prioritization but exhibited limitations, including static inputs and underestimation of outbreaks. Recommendations include recalibrating scoring, integrating real-time surveillance, enhancing seasonality modeling, and improving transparency (15–22). In Gabon, the STAR framework was applied to prioritize hazards. WHO’s Midterm Results Report (2024–2025) confirmed measles as a very high-risk hazard, prompting intensified immunization campaigns, stockpiling of emergency supplies, and enhanced surveillance. Cholera was assessed as very low risk, with preparedness focused on hygiene promotion and rapid response readiness, supported by WHO logistics hubs in Dakar and Nairobi. Meningitis was classified as high risk due to its seasonal nature and potential for cross-border transmission, necessitating cross-border coordination, vaccination campaigns, and contingency planning. These actions align with STAR’s emphasis on vulnerability, burden, and preparedness (18). In Guinea, 2024 outcomes validated STAR predictions. Measles experienced a major outbreak, confirming its high-risk classification and linking the resurgence to declining immunization coverage and post-pandemic disruptions. Cholera remained absent, consistent with its very low-risk prediction, suggesting effective preventive measures or unfavorable environmental conditions. Meningitis showed moderate activity, partially aligning with its high-risk classification, reinforcing the need for seasonal vaccination and cross-border coordination. These outcomes demonstrate STAR’s utility in hazard prioritization and resource allocation (19). In Mali, 2022 outcomes aligned closely with STAR predictions. Measles outbreaks confirmed its high-risk classification, linked to low immunization coverage, with only 70% of children receiving the first dose of the measles vaccine. Cholera did not occur, as predicted, matching its low-risk forecast, likely due to effective preventive measures and possibly favorable environmental conditions. Meningitis showed consistent presence with seasonal peaks, supporting its classification as a moderate to high-risk hazard and justifying continued vaccination campaigns and surveillance efforts (15). In Nigeria, statistical analysis revealed a strong correlation between predicted and actual cholera cases in 2024, indicating high model accuracy. Measles exhibited a moderate correlation but high variability, suggesting unpredictability in outbreaks, whereas meningitis lacked sufficient data for meaningful analysis. Recommendations included improving data quality, adopting advanced predictive models, and integrating external data sources such as climate and mobility patterns (20). Across all countries, several cross-cutting themes emerged. Meningitis predictions were generally accurate due to strong seasonality, whereas measles posed challenges due to variability and underprediction. Cholera predictions often overestimated risk or failed to capture localized outbreaks. The report emphasized the need for dynamic modeling, real-time data integration, and feedback loops to refine forecasting accuracy (21). The STAR tool itself was evaluated in detail. Its strengths include a structured risk assessment framework, support for prioritization, and multi-hazard capability. However, limitations were noted, such as underestimating outbreaks, reliance on static inputs, and limited geographic granularity. Recommendations for improving STAR included recalibrating scoring parameters, integrating real-time surveillance, enhancing seasonality modeling, and increasing transparency (22).

Overall, aligning predictive models with outbreak data is critical for preparedness. The report recommends implementing improved data systems, training in forecasting tools, and cross-sector collaboration to enhance early warning systems (15–22).